We present motivation and new Stata commands for modeling count data. While the focus of this article is on modeling data with underdispersion, the new command for fitting generalized Poisson regression models is also suitable as an alternative to negative binomial regression for overdispersed data.